A cluster is a group of independent servers that collaborate as a single system. The primary cluster components are processor nodes, a cluster interconnect (private network), and a disk subsystem. The clusters share disk access and resources that manage the data, but each distinct hardware cluster nodes do not share memory. Each node has its own dedicated system memory as well as its own operating system, database instance, and application software. Clusters can provide improved fault resilience and modular incremental system growth over single symmetric multi-processors systems. In the event of subsystem failures, clustering ensures high availability. Redundant hardware components, such as additional nodes, interconnects, and shared disks, provide higher availability. Such redundant hardware architectures avoid single points-of-failure and provide fault resilience.
In a database cluster, CPU and memory requirements for each node may vary depending on the database application. Performance and cost requirements also vary between database applications. One factor that contributes to performance is that each node in a cluster needs to keep other nodes in that cluster informed of its health and configuration. This has been done by periodically broadcasting a network message, called a heartbeat, across a network. The heartbeat signal is usually sent over a private network, a cluster interconnect, which is used for internode communications. However, lost or delayed heartbeat messages may cause false reports that a node is not functioning.
In prior systems, the cluster interconnect has been built by installing network cards in each node and connecting them by an appropriate network cable and configuring a software protocol to run across the wire. The interconnect was typically a low-cost/slow-speed Ethernet card running TCP/IP or UDP, or a high-cost/high-speed proprietary interconnect like Compaq's Memory Channel running Reliable DataGram (RDG) or Hewlett-Packard's Hyperfabric/2 with Hyper Messaging Protocol (HMP). A low-cost/high-speed interconnect would reduce clustering costs for users and reduce latency during run-time.